Summary: A Wall Street Journal report relayed by TechCrunch says a coalition of state attorneys general is investigating OpenAI, with New York’s attorney general serving a subpoena on June 12, 2026. The reported scope covers advertising practices, engagement and retention, model sycophancy, consumer and health data, and protections for minors and seniors. The news arrived four days after OpenAI confidentially filed IPO paperwork, and no timeline or full state list was disclosed.
A Wall Street Journal report, relayed by TechCrunch on June 13, describes a coalition of state attorneys general that has opened an investigation into OpenAI — with New York’s attorney general already serving the company a subpoena on Friday, June 12. No single state is going it alone here. This is coordinated enforcement across multiple states, and the subject list stretches from advertising practices to the protection of children.
What sharpens the story is when it happened. OpenAI had confidentially filed its IPO paperwork on June 8, a step behind Anthropic. Then, four days on, the subpoena landed — placing the company’s consumer-protection record under bright lights at the exact moment its path to going public begins.
What the Subpoena Reaches Into
The reporting lays out a scope that is at once wide and precise:
- Advertising practices
- User engagement and retention
- Model sycophancy
- Handling of consumer and health data
- Protections for minors and seniors
The list is itself the message. Decisions about product design — the mechanics of keeping ChatGPT users engaged — are being weighed as potential legal exposure, not brushed aside as engineering trivia. The standout entry is sycophancy: a model’s habit of flattering its users once lived comfortably in research papers, and it now sits squarely inside consumer-protection enforcement.
New York Out Front, the Rest of the Coalition Unnamed
One thing is solid: New York’s AG is part of this, at the very least as the office that delivered the subpoena. TechCrunch queried the New York attorney general’s office and heard nothing back, and OpenAI would not say which other states have joined. The investigation’s full outline remains unknown to the outside — a posture multi-state coalitions tend to hold on purpose during the early going.
For OpenAI, a roster kept hidden may cut deeper than one published in full. Consumer-protection statutes and litigation styles differ from state to state, which means bracing for several playbooks simultaneously — without knowing which are actually in play.
How OpenAI Answered
The company’s spokesperson reached for the familiar script: “AI is a new and powerful technology, and we work every day to safely bring its benefits to people in a responsible way,” followed by, “We take the concerns raised by state attorneys general seriously and intend to engage constructively with their offices.”
The minors question drew the fullest answer. ChatGPT as it stands, the company says, gives underage users a more protective experience, with safeguards that “direct them to real-world resources and trusted human contacts,” and points to age prediction, parental tools, and a prohibition on ads aimed at children as proof — closing on the words “We believe kids should be treated like kids.” Nothing in that answer is accidental; it is built to match the minors prong of the subpoena.
An IPO Filing Arrives Mid Legal Storm
Step back and the subpoena arrives inside the densest stretch of legal risk OpenAI has yet faced. On June 1, Florida attorney general James Uthmeier brought a first-of-its-kind suit against OpenAI and CEO Sam Altman, alleging they “ignored internal and external safety warnings,” “put children at great risk,” and “allowed a dangerous product to reach millions of Floridians.” Further back, November 2025 saw seven families file wrongful-death suits tying ChatGPT to suicides and delusions. Each episode feeds the same hardening regulatory mood that has run through 2026 from its opening weeks (see our 2026 AI opening outlook).
The IPO consequences run through the mechanics. Underwriting and disclosure demand a full account of material legal risks, and an investigation by a multi-state coalition is all but guaranteed a place in the filing’s risk-factors section.
The Takeaways for Product Teams
The practical effects come in threes. Interaction design in chatbots now sits directly inside consumer-protection enforcement: retention mechanics, notification strategy, and default behavior toward vulnerable users are all fair game for examination. Conversation data that borders on health is an explicit red line, and any product picking up on mental- or physical-health cues ought to audit its data flows. And a state-by-state regime means heavier compliance overhead — AI products sold into the US market increasingly have to plan compliance per state rather than at the federal level.
No clock has been set for the investigation to wrap up, but the trajectory is unmistakable: how persuasive and sticky an AI product is has stopped being a growth story and started being a legal-risk one.